车牌识别外文文献翻译中英文Word格式.docx
《车牌识别外文文献翻译中英文Word格式.docx》由会员分享,可在线阅读,更多相关《车牌识别外文文献翻译中英文Word格式.docx(17页珍藏版)》请在冰点文库上搜索。
2964-2968.
英文原文
LicensePlateRecognitionBasedOnPriorKnowledgeQianGao,XinnianWangandGongfuXie
Abstract-Inthispaper,anewalgorithmbasedonimprovedBP(backpropagation)neuralnetworkforChinesevehiclelicenseplaterecognition(LPR)isdescribed.Theproposedapproachprovidesasolutionforthevehiclelicenseplates(VLP)whichweredegradedseverely.Whatitremarkablydiffersfromthetraditionalmethodsistheapplicationofpriorknowledgeoflicenseplatetotheprocedureoflocation,segmentationandrecognition.Colorcollocationisusedtolocatethelicenseplateintheimage.Dimensionsofeachcharacterareconstant,whichisusedtosegmentthecharacterofVLPs.TheLayoutoftheChineseVLPisanimportantfeature,whichisusedtoconstructaclassifierforrecognizing.Theexperimentalresultsshowthattheimprovedalgorithmiseffectiveundertheconditionthatthelicenseplatesweredegradedseverely.
IndexTerms-Licenseplaterecognition,priorknowledge,vehicle
licenseplates,neuralnetwork.
I.INTRODUCTION
VehicleLicense-Plate(VLP)recognitionisaveryinterestingbutdifficultproblem.Itisimportantinanumberofapplicationssuchasweight-and-speed-limit,redtrafficinfringement,roadsurveysandparksecurity[1].VLPrecognitionsystemconsistsoftheplatelocation,thecharacterssegmentation,andthecharactersrecognition.Thesetasksbecomemoresophisticatedwhendealingwithplateimagestakeninvariousinclinedanglesorundervariouslighting,weatherconditionandcleanlinessoftheplate.Becausethisproblemisusuallyusedinreal-timesystems,itrequiresnotonlyaccuracybutalsofastprocessing.MostexistingVLPrecognitionmethods[2],[3],[4],[5]reducethecomplexityandincreasetherecognitionratebyusingsomespecificfeaturesoflocalVLPsandestablishingsomeconstrainsontheposition,distancefromthecameratovehicles,andtheinclinedangles.Inaddition,neuralnetworkwasusedtoincreasetherecognitionrate[6],[7]butthetraditionalrecognitionmethodsseldomconsiderthepriorknowledgeofthelocalVLPs.Inthispaper,weproposedanewimprovedlearningmethodofBPalgorithmbasedonspecificfeaturesofChineseVLPs.TheproposedalgorithmovercomesthelowspeedconvergenceofBPneuralnetwork[8]andremarkableincreasestherecognitionrateespeciallyundertheconditionthatthelicenseplateimagesweredegradeseverely.
II.SPECIFICFEATURESOFCHINESEVLPS
A.Dimensions
Accordingtotheguidelineforvehicleinspection[9],alllicenseplatesmustberectangularandhavethedimensionsandhaveall7characterswritteninasingleline.Underpracticalenvironments,thedistancefromthecameratovehiclesandtheinclinedanglesareconstant,soallcharactersofthelicenseplatehaveafixedwidth,andthedistancebetweenthemediumaxesoftwoadjoiningcharactersisfixedandtheratiobetweenwidthandheightisnearlyconstant.Thosefeaturescanbeusedtolocatetheplateandsegmenttheindividualcharacter.B.Colorcollocationoftheplate
TherearefourkindsofcolorcollocationfortheChinesevehiclelicenseplate.ThesecolorcollocationsareshownintableI.
TABLEI
Moreover,militaryvehicleandpolicewagonplatescontainaredcharacterwhichbelongstoaspecificcharacterset.Thisfeaturecanbeusedtoimprovetherecognitionrate.
C.LayoutoftheChineseVLPS
ThecriterionofthevehiclelicenseplatedefinesthecharacterslayoutofChineselicenseplate.AllstandardlicenseplatescontainChinesecharacters,numbersandletterswhichareshowninFig.1.ThefirstoneisaChinesecharacterwhichisanabbreviationofChinese
provinces.ThesecondoneisaletterrangingfromAtoZexcepttheletter
I.Thethirdandfourthonesarelettersornumbers.Thefifthtoseventhonesarenumbersrangingfrom0to9only.Howeverthefirstortheseventhonesmayberedcharactersinspecialplates(asshowninFig.1).Aftersegmentationprocesstheindividualcharacterisextracted.Takingadvantageofthelayoutandcolorcollocationpriorknowledge,theindividualcharacterwillenteroneoftheclasses:
abbreviationsofChineseprovincesset,lettersset,lettersornumbersset,numberset,specialcharactersset.
(a)Typicallayout
(b)Specialcharacter
Fig.1ThelayoutoftheChineselicenseplate
III.THEPROPOSEDALGORITHM
Thisalgorithmconsistsoffourmodules:
VLPlocation,charactersegmentation,characterclassificationandcharacterrecognition.ThemainstepsoftheflowchartofLPRsystemareshowninFig.2.
Firstlythelicenseplateislocatedinaninputimageandcharactersaresegmented.Theneveryindividualcharacterimageenterstheclassifiertodecidewhichclassitbelongsto,andfinallytheBPnetworkdecideswhichcharacterthecharacterimagerepresents.
A.Preprocessingthelicenseplate
1)VLPLocation
Thisprocesssufficientlyutilizesthecolorfeaturesuchascolorcollocation,colorcentersanddistributionintheplateregion,whicharedescribedinsectionII.Thesecolorfeaturescanbeusedtoeliminatethedisturbanceofthefakeplate’sregions.TheflowchartoftheplatelocationisshowninFig.3.
Fig.3Theflowchartoftheplatelocationalgorithm
Theregionswhichstructureandtexturesimilartothevehicleplateareextracted.Theprocessisdescribedasfollowed:
Here,theGaussianvarianceissettobelessthanW/3(Wisthecharacterstrokewidth),so1PgetsitsmaximumvalueMatthecenterofthestroke.Afterconvolution,binarizationisperformedaccordingtoathresholdwhichequalsT*M(T<
0.5).Medianfilterisusedtopreservetheedgegradientandeliminateisolatednoiseofthebinaryimage.AnN
*Nrectanglemedianfilterisset,andNrepresentstheoddintegermostlyclosetoW.
Morphologyclosingoperationcanbeusedtoextractthecandidateregion.Theconfidencedegreeofcandidateregionforbeingalicenseplateisverifiedaccordingtotheaspectratioandareas.Here,theaspectratioissetbetween1.5and4forthereasonofinclination.Thepriorknowledgeofcolorcollocationisusedtolocateplateregionexactly.ThelocatingprocessofthelicenseplateisshowninFig.4.
2)Charactersegmentation
Thispartpresentsanalgorithmforcharactersegmentationbasedonpriorknowledge,usingcharacterwidth,fixednumberofcharacters,theratioofheighttowidthofacharacter,andsoon.TheflowchartofthecharactersegmentationisshowninFig.5.
Firstly,preprocessthelicensetheplateimage,suchasunevenilluminationcorrection,contrastenhancement,inclinecorrectionandedgeenhancementoperations;
secondly,eliminatingspacemarkwhichappearsbetweenthesecondcharacterandthethirdcharacter;
thirdly,mergingthesegmentedfragmentsofthecharacters.InChina,allstandardlicenseplatescontainonly7characters(seeFig.1).Ifthenumberofsegmentedcharactersislargerthanseven,themergingprocessmustbeperformed.TableIIshowsthemergingprocess.Finally,extractingtheindividualcharacter’imagebasedonthenumberandthewidthofthecharacter.Fig.6showsthesegmentationresults.(a)Theinclineandbrokenplateimage,(b)theinclineanddistortplateimage,(c)theseriousfadeplateimage,(d)thesmutlicenseplateimage.
whereNfisthenumberofcharactersegments,MaxFisthenumberofthelicenseplate,andiistheindexofeachcharactersegment.
Themediumpointofeachsegmentedcharacterisdeterminedby:
(3)
where1iS
istheinitialcoordinatesforthecharactersegment,and2iSisthe
finalcoordinateforthecharactersegment.Thedistancebetweentwoconsecutivemediumpointsiscalculatedby:
(4)
Fig.6Thesegmentationresults
B.Usingspecificpriorknowledgeforrecognition
ThelayoutoftheChineseVLPisanimportantfeature(asdescribedinthesectionII),whichcanbeusedtoconstructaclassifierforrecognizing.Therecognizingprocedureadoptedconjugategradientdescentfastlearningmethod,whichisanimprovedlearningmethodofBPneuralnetwork[10].Conjugategradientdescent,whichemploysaseriesoflinesearchesinweightorparameterspace.Onepicksthefirstdescentdirectionandmovesalongthatdirectionuntiltheminimuminerrorisreached.Theseconddescentdirectionisthencomputed:
thisdirectionthe“conjugatedirection”istheonealongwhichthegradientdoesnotchangeitsdirectionwillnot“spoil”thecontributionfromthepreviousdescentiterations.Thisalgorithmadoptedtopology625-35-NasshowninFig.7.Thesizeofinputvalueis625(25*25)andinitialweightsarewithrandomvalues,desiredoutputvalueshavethesamefeaturewiththeinputvalues.
AsFig.7shows,thereisathree-layernetworkwhichcontainsworkingsignalfeedforwardoperationandreversepropagationoferrorprocesses.Thetargetparameteristandthelengthofnetworkoutput
vectorsisn.Sigmoidisthenonlineartransferfunction,weightsareinitializedwithrandomvalues,andchangedinadirectionthatwillreducetheerrors.
Thealgorithmwastrainedwith1000imagesofdifferentbackgroundandilluminationmostofwhichweredegradeseverely.Afterpreprocessingprocess,theindividualcharactersarestored.Allcharactersusedfortrainingandtestinghavethesamesize(25*25).Theintegratedprocessforlicenseplaterecognitionconsistsofthefollowingsteps:
1)Featureextracting
Thefeaturevectorsfromseparatedcharacterimageshavedirecteffectsontherecognitionrate.Manymethodscanbeusedtoextractfeatureof